Vibe Coding DiscoverCNEquity
View on GitHub中国 A 股数据基础设施。42 个日更数据集：行情、基本面、资金面、公告事件、指数行业、宏观与风险。行级溯源、PIT 语义、复权与历史成分内置，MCP 原生。自托管，零注册、零 API Token
Self-hosted China A-share data lake with 42 daily-updated datasets in Parquet/DuckDB, row-level provenance, PIT semantics and adjustment built in. Ships 6 native MCP tools so AI agents and Python/Polars/DuckDB clients query the same local lake, with no registration or API token.
Use Cases
Query China A-share data from an AI agent over MCP toolsBacktest strategies on a local point-in-time data lake without re-fetching dataLoad adjusted (hfq/qfq) daily bars and valuation metrics via Python/Polars/DuckDBAvoid survivorship bias by keeping delisted stocks and historical index constituentsPoint-in-time fundamental and announcement queries for factor researchSelf-host daily-updated market, fund-flow, macro and sentiment datasets with no API tokenRun scheduled incremental daily ingestion and coverage/staleness auditsTrace each row back to its source, data version and fetch time
Built With
- Language
- Python
- Frameworks
- MCP · DuckDB · Polars · PyArrow · Parquet · Click · pytest
Tags
a-shares · china-stock-market · financial-data · mcp · mcp-server · data-lake · duckdb · parquet · polars · quantitative-finance · backtesting · point-in-time · market-data · self-hosted · llm-tools · python